CRAT: A Multi-Agent Framework for Causality-Enhanced Reflective and Retrieval-Augmented Translation with Large Language Models

Meiqi Chen,Fandong Meng,Yingxue Zhang,Yan Zhang,Jie Zhou

2024-10-28

Abstract:Large language models (LLMs) have shown great promise in machine translation, but they still struggle with contextually dependent terms, such as new or domain-specific words. This leads to inconsistencies and errors that are difficult to address. Existing solutions often depend on manual identification of such terms, which is impractical given the complexity and evolving nature of language. While Retrieval-Augmented Generation (RAG) could provide some assistance, its application to translation is limited by issues such as hallucinations from information overload. In this paper, we propose CRAT, a novel multi-agent translation framework that leverages RAG and causality-enhanced self-reflection to address these challenges. This framework consists of several specialized agents: the Unknown Terms Identification agent detects unknown terms within the context, the Knowledge Graph (KG) Constructor agent extracts relevant internal knowledge about these terms and retrieves bilingual information from external sources, the Causality-enhanced Judge agent validates the accuracy of the information, and the Translator agent incorporates the refined information into the final output. This automated process allows for more precise and consistent handling of key terms during translation. Our results show that CRAT significantly improves translation accuracy, particularly in handling context-sensitive terms and emerging vocabulary.

Computation and Language

What problem does this paper attempt to address?

### The Problem the Paper Attempts to Solve

This paper aims to address the issues of inconsistency and errors in handling context-dependent terms encountered by large language models (LLMs) in machine translation. Specifically:

1. **Handling of Context-Dependent Terms**: LLMs tend to exhibit inconsistencies and errors when dealing with new words or domain-specific terms. These terms may have different meanings in different contexts, leading to inaccurate translation results.

2. **Limitations of Manual Term Identification**: Existing solutions often rely on manual identification of these terms, but this approach is impractical in real-world applications. The complexity and ever-evolving nature of language make manual identification difficult to achieve.

3. **Hallucinations Caused by Information Overload**: Although retrieval-augmented generation (RAG) can provide some assistance, its application in translation is limited by the hallucination problem caused by information overload. Excessive external information can interfere with accurate translation, leading to errors.

To address these issues, the paper proposes CRAT (Causally Reinforced Reflective and Retrieval-Augmented Translation Multi-Agent Framework). This framework improves translation consistency and accuracy by automatically identifying unknown terms, extracting internal and external knowledge, verifying the accuracy of information, and ultimately generating precise translations.